A public chatbot may receive prohibited topics and personal information.
Apply policy to both inputs and outputs.
Which operational design is most appropriate?
Combine input/output checks, policy rules, monitoring, minimized logs, and human escalation for exceptional cases.
Detailed explanation
Layered controls are more robust than relying on one prompt instruction.
Layered controls are more robust than relying on one prompt instruction.
A prompt alone cannot guarantee safe behavior for every input and output.
A prompt alone cannot guarantee safe behavior for every input and output.
Logs also require minimization, access control, and retention design.
Logs also require minimization, access control, and retention design.
Without measurement, controls cannot be verified or improved.
Without measurement, controls cannot be verified or improved.
Try it yourself
An example you can run in a temporary verification environment.
Amazon Bedrock Guardrails公式ドキュメントとAWS Well-Architectedの安全性設計を確認する。Expected result
安全制御をポリシー・検査・監視・人手対応の多層構成で説明できる。Key points
- Input and output controls
- Monitoring
- Escalation
Notes
- Environment: AWS公式AIF-C01試験ガイドとAWS公式ドキュメントの確認
- Command output formatting can vary slightly by distribution or tool version.
- Run the example in a temporary directory or process when possible.
Foundation review
Read the scope first
Check whether the command acts on the current shell, a new process, an existing process, or a file.
Verify the observable result
Use the supplied command and compare the output with the expected result.